How to run qualitative research

Conversion audits flag leaks, yet numbers alone rarely tell you why visitors hesitate. Qualitative research fills that gap. I discovered this while optimising a pricing page that looked fine in analytics but left prospects cold. A call with three recent buyers revealed confusion over hidden fees. A quick copy tweak lifted form submissions by thirty per cent within a week.

This chapter shows how to extract that kind of insight. You will interview customers, analyse on-page behaviour with heatmaps, run lightweight surveys and turn every finding into a growth backlog. Use the steps in order and each sprint will start with problems worth solving, not random hypotheses.

Begin with customer conversations. Schedule five twenty-minute calls with people who booked a meeting in the last month. Ask open questions: “What nearly stopped you from booking?” and “Which line made you trust us?” Record the calls and let tools such as Notion AI, Fireflies or Google Meet transcript extract key phrases automatically.

Highlight exact wording. Phrases like “felt risky” or “saved me hours” become headline gold later. Ignore small sample size worries; repeated phrases across five calls usually signal a

Install a heatmap tool such as Hotjar or Microsoft Clarity. Run recordings for one thousand unique visits or seven days. Focus on high-value pages like pricing, case studies and contact forms. Look for rage clicks, scroll drop-offs and hesitation pauses longer than three seconds.

Overlay findings with the motivator and friction lists. If recordings show repeated hover on feature FAQs, surface that content higher on the page. When users scroll past social proof, tighten header copy so proof appears sooner.

Capture screenshots of standout sessions and annotate them. Visual evidence speeds stakeholder buy-in during backlog grooming.

Heatmaps give macro patterns, but you still need quick quantitative checks. Surveys supply those signals next.

Launch an exit-intent poll on key pages. Ask one question only: “What stopped you from booking a call today?” Offer predefined answers plus an “Other” box. Keep the poll live until you collect one hundred responses to ensure significance.

Pair this with a post-demo survey emailed to booked leads. Ask “What nearly stopped you from booking?” and “Which part convinced you?” Compare answers with exit poll data. Overlapping frictions jump straight to the top of the backlog. Divergent answers suggest segmented messaging is required.

Plot responses on a simple impact versus frequency grid. High-impact, high-frequency issues become immediate test themes.

With data in hand, you need a place to store and prioritise ideas. That structure forms the final section.

Create a growth backlog in Notion or Trello. Each card contains the insight source, problem statement, hypothesis and a rough estimate of lift potential. Tag cards by page and funnel stage. Assign confidence scores based on the number of qualitative sources that support the hypothesis.

Rank cards using the ICE framework: impact, confidence and ease. Review top five cards every sprint planning session and assign at least one to development or copy updates. Mark implemented ideas with results once metrics roll in. This feedback loop prevents forgotten learnings and avoids repeated mistakes.

Share the backlog link with marketing, product and design teams. Transparent priorities align efforts and speed sign-offs.

The backlog closes the research loop and feeds the experimentation phase that follows in the guide.

Qualitative research turns anonymised clicks into human stories. Customer interviews reveal language and fears. Heatmaps expose hidden friction. Surveys quantify objections at scale. A structured backlog converts every insight into testable actions.

Run this cycle before building any A/B variant. You will enter experimentation armed with real problems and buyer language that short-circuits guesswork. Tiny lifts will stack, sprint by sprint, into a conversion engine that doubles booked meetings within the year.